Polyploidy and Discontinuous Heredity Effect on Evolutionary Multi-Objective Optimization
This work addresses optimization problems in fields like engineering or data science by introducing a biologically inspired method, though it appears incremental as it builds on existing evolutionary algorithms.
The paper tackles the challenge of improving evolutionary multi-objective optimization by mimicking discontinuous heredity from polyploid organisms, where traits can be silently carried and reappear, and finds that this representation outperforms the Non-Dominated Sorting Genetic Algorithm-II on benchmark problems with high decision variables and objectives.
This paper examines the effect of mimicking discontinuous heredity caused by carrying more than one chromosome in some living organisms cells in Evolutionary Multi-Objective Optimization algorithms. In this representation, the phenotype may not fully reflect the genotype. By doing so we are mimicking living organisms inheritance mechanism, where traits may be silently carried for many generations to reappear later. Representations with different number of chromosomes in each solution vector are tested on different benchmark problems with high number of decision variables and objectives. A comparison with Non-Dominated Sorting Genetic Algorithm-II is done on all problems.